Meta has parted ways with the co-founders and several employees of AI security startup Virtue AI, only four months after bringing the team into its Meta Superintelligence Labs in June 2026. The group included co-founders Bo Li, Dawn Song and Sanmi Koyejo.
Meta spokesperson Andy Stone confirmed the departures, saying the arrangement “didn’t work out as planned” because of differences in working styles. The Virtue AI team had been brought in to strengthen Meta's work around AI safety, alignment and frontier risks.
Virtue AI had focused on AI security, including automated red teaming, runtime guardrails and AI governance. Before joining Meta, the startup had worked with organisations including Anthropic, OpenAI and the U.S. National Institute of Standards and Technology (NIST).
The move comes as Meta continues to reshape its AI organisation and increase its focus on building advanced AI systems. Despite the departure of the Virtue AI group, Stone said Meta Superintelligence Labs remains focused on AI safety, alignment and frontier risk.
Why it matters: The unusually short four-month tenure highlights the challenges large technology companies can face when integrating specialised AI teams and startups into rapidly evolving internal AI organisations. It also shows how quickly AI talent strategies can change as companies compete to build and secure advanced AI systems.